[tests] infrastructure for automatic pytest of babs-submit and babs-status

[zhao-cy]

Currently we are trying to figure out feasible infrastructure for testing out babs-submit and babs-status automatically. These two programs require interactions w/ the clusters. We hope to be able to run pytest of babs-submit and babs-status in an automatical and controlled way and even on CircleCI.

Current ways we could think of:

1. emulator of cluster job scheduler system.

• Challenge 1: find one for SGE and one for Slurm.
• Challenge 2: install everything, including this emulator, and also singularity, python, datalad, etc in one Docker image, so that we can test it on CircleCI.
• Challenge 3: how to control this emulator to get the job status we want?

2. Just to use current real clusters.

• Challenge 1: how to control our clusters to get the job status we want?

3. job scheduler "replier": a simple Python package that returns printed messages we usually get from qstat (SGE) or squeue (Slurm) when checking job status.

• Challenge 1: we implement it (it might turn out to be not that complex!)
• Advantage 1: We have full control on the messages it returns / job statuses
• Advantage 2: We can design different versions of SGE or Slurm clusters, to add support for a specific cluster which has specific returned messages